Pedro – Python, Django, React

A tech pharma company that leverages trial development and analysis using AI.

• Optimized Asynchronous Task Processing: Enhanced Celery performance by properly managing I/O-bound operations asynchronously;
• Developed RAG Service: Built a retrieval-augmented generation (RAG) service for drug study generation using LangChain and pgvector;
• Streamlined Documentation: Created a new documentation platform with Starlight, improving team collaboration and onboarding;
• API Development: Designed GraphQL APIs using Strawberry and Django, supporting complex data interactions;
• Front-End Engineering: Developed React components with TypeScript and implemented comprehensive unit tests with Jest;
• Integration Testing: Established robust test suites with pytest for backend services;
• Wrote integration tests with pytest;
• Database Optimization: Resolved PostgreSQL performance issues and optimized vector search using IVFFlat indexes;
• Optimized vector searches by using the IVFFlat index;
• Saved costs by optimizing the Github Actions pipeline;
• Data Ingestion Services: Built high-throughput data ingestion pipelines from Snowflake Data Warehouse.

Notable Projects

• Automated Study Generation: Created a new RAG-based workflow for generating study suggestions, streamlining research processes.
• End-to-End Testing: Introduced Cypress for E2E testing, ensuring critical front-end features remain reliable in production.
• Documentation Culture: Improved the documentation workflow, encouraging consistent and thorough technical writing across the team.
• Data Science Integration: Built a complex service for trial feasibility analysis, leveraging data science techniques for large-scale data manipulation.

An intelligent service that automates and enhances the Bill of Materials (BOM) calculation process for construction projects. By combining Optical Character Recognition (OCR) to read CAD files with Retrieval-Augmented Generation (RAG), it automatically identifies building components and accurately calculates required construction materials. This streamlines the traditionally manual and time-consuming process of generating BOMs, helping construction professionals make faster and more accurate material estimations.

• Designed a software that uses OCR and LLM to read, measure, and calculate materials for civil constructions;
• Designed an API with Django;
• Modeled the database schema;
• Configured background tasks with Celery;
• Built workflows similar to RAG;
• Handled OCR data extraction from images;
• Manipulated DWG files with Python to extract dimensions from objects;
• Designed the entire UI with React, Shadcn, and TailwindCSS;
• Integrated Autodesk API to manage different use cases inside the application.

A local news platform providing an online dashboard that helps its users create and share content through email and their blog. It also handles paid subscriptions and advertisements.

• Built with my team a new dashboard for handling redirects at Cloudflare for legacy newsletters pages;
• Improved back-end and front-end architecture;
• Added internal libraries for Go applications to be shared among the microservices;
• Updated a Go service to use Recurly for managing subscriptions;
• Refined Algolia full-text search functionality in a Go monolith app;
• Contributed to documentation by writing ADRs for every improvement made;
• Improved database transactions;
• Wrote integration tests for Go services;
• Participated and discussed several features' implementation and their technical aspects;
• Responsible for introducing Postgres transactions in their Go monolith service;
• Solved a complex problem like Go transaction issues during integration tests.

It's Latin America's largest data marketplace, providing companies with insights into their audience and managing digital marketing campaigns all in one place.

• Worked on a tag manager to collect user information across different websites;
• Led a team of 4 developers to build an MVP for crossing user’s information among different websites;
• Migrated the monolith architecture to a microservice architecture;
• Automated integrations and deployments by creating CI/CD Github action pipelines;
• Conducted dailies, retros, and sprint planning;
• Approached different solutions for optimizing large datasets;
• Handled several integrations with 3rd party Adversments APIs such as Google Adsense;
• Configured a Kubernetes cluster and created the services files for our microservices.

It's an internal project developed to monitor and manage a pipe responsible for extracting oil from oceans.

• Worked with a nuclear physicist to create a program capable of calculating the probability of failure on demand when a BOP is operating;
• Built a feature to manage BOP campaigns;
• Helped design the database and improved indexing;
• Enhanced UI utilizing Vue.js;
• Improved mathematical calculus using Python and NumPy library;
• Customized charts library in JavaScript to adapt to our customer's needs.